.split() output seems not recognised by .transpose()

[ktpolanski]

Bug report

Expected behavior and actual behavior

Performing a .map() which includes a .split() on one of the it components seems to create a form of data that does not get picked up and expanded by a subsequent .transpose().

Steps to reproduce the problem

ch1 = Channel.of(["A", ["B", "C"]])
ch1.transpose().view()

ch2 = Channel.of(["D", "E_F"])
ch2.map({it -> [it[0], it[1].split("_")]}).transpose().view()

Program output

N E X T F L O W  ~  version 23.04.1
Launching `transpose-test.nf` [scruffy_hodgkin] DSL2 - revision: 958c877451
[D, [E, F]]
[A, B]
[A, C]

Environment

• Nextflow version: 23.04.1
• Java version: openjdk 17.0.6
• Operating system: Linux
• Bash version: GNU bash, version 4.4.20(1)-release (x86_64-pc-linux-gnu)

[ktpolanski]

Of note, adding .toList() as part of the .map() seems to resolve this situation:

ch2 = Channel.of(["D", "E_F"])
ch2.map({it -> [it[0], it[1].split("_").toList()]}).transpose().view()

yields

N E X T F L O W  ~  version 23.04.1
Launching `transpose-test.nf` [insane_mirzakhani] DSL2 - revision: 28f818061d
[D, E]
[D, F]

[mribeirodantas]

This is not a bug, as you probably noticed with your second comment. The split function returns a string that may look like a list with two strings, but it's a full string. You can confirm this with the snippet below:

ch1 = Channel.of(["A", ["B", "C"]])
ch1.map { println it[1].getClass() }

ch2 = Channel.of(["D", "E_F"])
ch2.map { it -> [it[0], it[1].split("_")] }.map { println it[1].getClass() }

The output of the snippet above is:

N E X T F L O W  ~  version 23.06.0-edge
Launching `foo.nf` [agitated_hilbert] DSL2 - revision: d5e6b1ed25
class java.util.ArrayList
class [Ljava.lang.String;

If you apply a toList to this string, then you have a list as the first case and things work accordingly :-)

[bentsherman]

Hi @ktpolanski , the split() method returns a string "array" (String[]) which is why you have to convert it to a list with toList(). Alternatively, you can use the tokenize() method which returns a regular list.

[ktpolanski]

tokenize() and split() technically behave a bit differently, potential users should be aware of this

ch2 = Channel.of("hello world")
ch2.map({it -> it.tokenize("lo")}).view()
ch2.map({it -> it.split("lo")}).view()

N E X T F L O W  ~  version 23.04.1
Launching `test.nf` [condescending_visvesvaraya] DSL2 - revision: d1db60d57a
[hel,  world]
[he,  w, r, d]